Prediction of presynaptic and postsynaptic neurotoxins by bi-layer support vector machine with multi-features

Much benefit to biology research and drug design, prediction of neurotoxin graduallybecame a necessary and popular task in recent year. In this paper, based on multi-feature extraction strategies from primary sequences and support vector machine, a novel Multi-classifier system named bi-layer support vector machine was proposed to predictpresynaptic and postsynaptic neurotoxins, and obtained satisfactory results with 98.5%prediction accuracies for presynaptic neurotoxins and 99.18% for postsynaptic neurotoxins, the Matthew’s correlation coefficient was 0.9767. The satisfactory results showed that, the current method might play a complementary role to other existing methods for predicting presynaptic and postsynaptic neurotoxins.   Key words: Prediction, bi-layer support vector machine, pseudo amino acid composition, approximate entropy, dipeptide.